Strategic issues in 1559
ETHTaipei, October 15th, 2020
Barnabé Monnot, Ethereum Foundation
The transaction lifecycle
Fee variance
In general, difficult to find an “entry price” into a block.
This is because users have to guess the “correct” price (or one that will get them in).
Basefee
1559 introduces a dynamic minimum gas price, basefee.
Parameters
Users specify two parameters:
- A premium, the maximum tip the miner receives from the transaction.
- A fee cap, the highest price the user is willing to pay.
Basefee dynamics
We start basefee at 1 Gwei.
More users come in than the chain can accommodate.
Users have different values for the transaction.
After price discovery, only high-value users get in.
Strategic users
An often-heard criticism:�“Basefee doesn’t change anything, users still compete with each other to get in!”
When is that true?
Answer: during price discovery.
Strategic users
Eventually, strategic users become price-takers again.
- Either they don’t value their transaction enough,
- Or they don’t need to compete with tip.
Learning users
“Next-block inclusion” is a marketed feature of 1559.
Current system: inclusion is not predictable.
We show that users:
- Either learn to expect next-block inclusion�- Or do not value the tx enough.
Learning users in 1559 (unpublished)
Before learning
After learning
Takeaways
The basefee operates price discovery for the transaction fee market.
In normal conditions, it allows for price-taking users.
In fact, users with high enough value learn to expect next-block inclusion.
When demand shifts, there is an incentive to be strategic, but this disappears once price discovery is over.
Some reasons to expect that demand shifts are quickly matched by the basefee, and happen infrequently enough.
Transition
Currently, consensus over “soft” transition period.
Users and apps can still use legacy transactions.
They will be cast into 1559-style transactions, and you are most likely losing from the cast.
Better to switch to 1559!
Genesis
First described in Vitalik’s Blockchain Resource Pricing (August 2018)
Formalised into EIP 1559 by Vitalik and Eric Conner (April 2019)
Working groups started by Alexey Akhunov and Rick Dudley (April 2019 - ...)
That one long Ethereum Magicians thread
What’s next
- Robust Incentives Group: More simulations investigating transition period towards EIP1559; floating escalator; wallet defaults; transaction pool policies.
- 1559 devs: Tim Beiko’s notes; 1559 Mainnet Readiness Checklist; testnets running with Besu and Nethermind (soon Geth fork by Vulcanize).
- Tim Roughgarden: Report on miner incentives under 1559 and framing fee market improvements in algorithmic game theory terms.
Contributions welcome! DMs are open.
Beating price discovery
Escalator: a different proposal by Dan Finlay.
Bid escalates over time. The faster it escalates ⇔ the faster it gets in.
Possible to combine 1559 and escalator: “floating escalator”.